Jump to content

Recommended Posts

Posted (edited)

So, SATA... really? Makes me feel like I'm configuring a desktop... :(

TPM

UEFI instead of bios (apparantly for disks over 2TB)? I've never used it that I know of, any issues?

Partition - GPT enabled (again a large disk thing)?

RDIMM over UDIMM (I've read RDIMM is preferable but I can only configure this for 1x16GB stick and I'd rather have 2x8GB obviously)?

Dual redundant hot-plug power supply is it really worth it?

 

Thanks,

 

Kol.

 

OK:

 

1) What do you see yourself doing with these servers? What is more important to you, disk throughput or capacity? For what it's worth, we have SATA disks in a lot of our servers and even in a couple of our SANs

2) The TPM can be used with Bitlocker to encrypt your disk volumes. Handy if the server gets pinched, it means they can't take the disks out, put them in another machine and copy the data.

3) You can't boot from a volume greater than 2TB on a BIOS. The problem with UEFI is that you need Windows 2012 to properly support it

4) Again, volumes larger than 2TB need to be on a GPT disk

5) I forget which way round it is but you can put higher amounts of RAM in with one type than you can with the other

6) Yes, yes, a thousand times yes. It means that if a PSU does go, you can still run the school while you're getting a replacement. It's even more important if you're just running one server.

Edited by Norphy
  • Thanks 1
Posted

Well, the purpose of the server isn't fully envisaged yet, only that the current one needs to be replaced.

 

For the future I'm hoping to use it as 'my baby' and put SCCM, WSUS, MDT and images on it, possibly remote access. Maybe something like Impero, but basically have it as the server for controlling the environment. Maybe even looking at VM's in the future. Hopefully it will be the main server for backups and software deployment and have all the cool toys in one place without upsetting the school 'production environment' while I'm working on things and testing or if an install of any of the above goes bad. Also it will be taking the 'roles'... As for school data, SIMS, print server and such that will all be done from the other server, but they will both be DC's as I'm nervous about only having one. Anyway, suffice to say 'I' have big plans for this server and I'd like it to be able to be enough to cope with at least some future IT needs...

 

I really can't see anyone managing to make off with the server, it's in a locked room and there are far more 'shiny' things around for the opportunistic thief.

 

I'm not willing to jump into 2012 as yet, I'd rather it was a more mature product and as I'm going to have a lot to configure I'd rather it be with an OS I'm a little more familiar with. Is there really any actual 'problem' with UEFI and 2008R2? I'm not sure what amazing new things 2012 brings that make the effort worth it...

 

So, I think the disks are 1TB per disk but will be seen by the OS as 2TB as there will be 4 in RAID 10. So I'm not sure GPT will be strictly necessary, but maybe for future proofing it will be worth it.

 

This means a few things, UDIMM is limited to two DIMMs per memory channel and UDIMMs offer slightly better memory bandwidth for one DIMM per channel. However when using two or three DIMMS per channel you will get better memory bandwidth with RDIMMs. This is because when you go to 2 DIMMs per memory channel, due to the high electrical loading on the address and control lines, the memory controller uses a “2T” or “2N” timing for UDIMMs. Consequently every command that normally takes a single clock cycle is stretched to two clock cycles to allow for settling time. Therefore, for two or more DIMMs per channel, RDIMMs will have lower latency and better bandwidth than UDIMMs.

 

So I take it this means I have to know now, whether I will want to put more memory in at a later date? As it seems UDIMMs will limit putting more than 2 in, or get big fat UDIMMs now and hope that will last me. Maybe the limitation isn't the same now, but I seem to remember that large memory modules (over 512MB back in the day lol) should be registered? Think I am going to go RDIMM if possible but I also want more than one big stick of RAM (as in the spec above) in case it goes bad.

 

Ok, ok redundant PSU it is then!

 

Kol.

Posted (edited)
For the future I'm hoping to use it as 'my baby' and put SCCM, WSUS, MDT and images on it, possibly remote access. Maybe something like Impero, but basically have it as the server for controlling the environment. Maybe even looking at VM's in the future. Hopefully it will be the main server for backups and software deployment and have all the cool toys in one place without upsetting the school 'production environment' while I'm working on things and testing or if an install of any of the above goes bad. Also it will be taking the 'roles'... As for school data, SIMS, print server and such that will all be done from the other server, but they will both be DC's as I'm nervous about only having one. Anyway, suffice to say 'I' have big plans for this server and I'd like it to be able to be enough to cope with at least some future IT needs...

 

All of that on one server is a bad, bad idea. SCCM in itself is very heavy and ideally wants to live on its own hardware. It needs its own SQL server which is very RAM and disk intensive. Deploying software and OS images are I/O intensive in themselves and if they're coming from the same server that's serving files and your SIMS database...

 

Not good.

Edited by Norphy
Posted
I'm not willing to jump into 2012 as yet, I'd rather it was a more mature product and as I'm going to have a lot to configure I'd rather it be with an OS I'm a little more familiar with. Is there really any actual 'problem' with UEFI and 2008R2? I'm not sure what amazing new things 2012 brings that make the effort worth it...

 

Usually, I'd agree with that, but 2012 is superb. There is now 2012 R2 also, which is even better. Just by using it, you'll get more out of the hardware. For example, disk IO is more efficient in 2012.

Posted
2012 isn't that different either. Once you get away from the start menu, most things live in the same place as they've always done. The ones that don't live in Server Manager. It's not a huge leap in terms of interface but there are some benefits such as dedup. The space savings we have on our servers through that are huge.
Posted
All of that on one server is a bad, bad idea. SCCM in itself is very heavy and ideally wants to live on its own hardware. It needs its own SQL server which is very RAM and disk intensive. Deploying software and OS images are I/O intensive in themselves and if they're coming from the same server that's serving files and your SIMS database...

 

Not good.

 

For the future I'm hoping to use it as 'my baby' and put SCCM, WSUS, MDT and images on it, possibly remote access. Maybe something like Impero, but basically have it as the server for controlling the environment. Maybe even looking at VM's in the future. Hopefully it will be the main server for backups and software deployment and have all the cool toys in one place without upsetting the school 'production environment' while I'm working on things and testing or if an install of any of the above goes bad. Also it will be taking the 'roles'... As for school data, SIMS, print server and such that will all be done from the other server, but they will both be DC's as I'm nervous about only having one. Anyway, suffice to say 'I' have big plans for this server and I'd like it to be able to be enough to cope with at least some future IT needs...

 

Sorry, wall of text might have made that a TL;DR... it's the 'other' server that holds SIMS, serves files and does print management. This new server will be the one for SCCM etc... Hence the reason for wanting RDIMMS and SCSI drives. :D

 

Kol.

Posted

Ah right, sorry, wasn't a tl;dr, just misread on my part.

 

Don't worry too much about the speed of your RAM, the amount of it is more important. It's better to have 32GB of slower RAM than 16GB of fast RAM, especially on a box hosting SQL.

 

Good IO throughput is important but so is overall capacity. I have just under a terabyte of storage on my ConfigMgr server and I have just under 76GB left on the volume that's holding my DP. I want more storage on there but I don't know if I'll be able to get it. Get some big SAS (600GB minimum) drives in there if you can get the budget but don't stress too much about putting SATA ones in there if you end up having to.

 

Putting SCCM on a DC is not recommended however.

  • Thanks 1
Guest Guest
Posted

One thing which has limited me when repurposing servers when they become EOL in their primary role is hard drive bay number. I tend to use whichever old server is the best and use it as a storage or backup box. With that in mind I'd upgrade the front of the chassis to accept the largest number of hard drives possible, even if you don't plan on using the capacity now.

 

An extra £100 or so now saves you from having the buy a whole server just for backups in the future.

Posted
Running WDS/WSUS/Software Deploy/Printer etc, doesn't really use that much processing, unless you're talking thousands of machines.

 

SCCM in itself is very heavy and ideally wants to live on its own hardware. It needs its own SQL server which is very RAM and disk intensive. Deploying software and OS images are I/O intensive in themselves...

 

Hmmm... Well for every point of view there is an opposite I guess, it's all relative and down to each individual experience.

 

Putting SCCM on a DC is not recommended however.

 

Thing is it's getting more and more like that, but in reality I can't put SIMS on it's own server (as has been suggested somewhere), SCCM on it's own server, have another for this, another for that...

 

It will be a DC because we will only ever have 2 servers and I just don't like the idea of only having 1 DC with no replication/back-up on-line, working and ready to step in. I mean I guess I could have some much smaller servers and more of them, or go the virtual route (although that's not giving them their own hardware) and I'm sure hardware failure taking down 2 or 3 VM's as well as the host is easier to reinstate, but it's a whole other (practically vertical) learning curve for me to deal with.

 

Kol.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...